Path: chuka.playstation.co.uk!news
From: gil@snsys.com (Gil Jaysmith)
Newsgroups: scea.yaroze.freetalk
Subject: Re: C versus C++
Date: Mon, 19 Jan 1998 11:15:58 GMT
Organization: PlayStation Net Yaroze (SCEE)
Lines: 13
Message-ID: <34c33527.852132@www-s.playstation.co.uk>
References: <69cmja$7pn1@emeka.playstation.co.uk> <34b9e34f.14105854@news.scea.sony.com> <34BF10FF.13D7@bc.sympatico.ca> <34C293DF.6237DFF6@forefront.com.au>
NNTP-Posting-Host: trish.snsys.com
X-Newsreader: Forte Free Agent 1.11/32.235

On Mon, 19 Jan 1998 10:44:31 +1100, Toby Sargeant
<toby@forefront.com.au> wrote:
>Current wisdom seems to be that the GCC codegen is really rather tight.

The per-function code generation of GCC is okay (provided you have
lots of virtual memory). It's got bugger-all cross-function
optimisation and the code motion isn't up to much. It also sometimes
doesn't reoptimise the assembly if it breaks up single insns into
multiple ones in the hope that it can then peephole subsequent
instructions and weave them in; it should reform the original
instruction. This happens primarily with 32-bit register loads.

- Gil